Alibaba's Qwen4-Max Scores 43, 16 Points Behind Claude Opus 5.5
Claude Fable 5.2 scored 65, putting it 22 points ahead of Alibaba's flagship model.

Alibaba's Qwen4-Max scored 43 on an AI intelligence index after its Nov. 16 launch, 16 points below Anthropic's Claude Opus 5.5.
Claude Fable 5.2 scored 65, 22 points above Qwen4-Max. The results place Alibaba's flagship model behind both Claude models on the index.
Opus 5.5 scored 59 on the index, six points below Fable 5.2.
The score gap alone does not show why Qwen4-Max ranked below the Claude models.
Alibaba said in September that it was training Qwen4 on a new architecture.
The company had also listed Singapore rates for Qwen4-Max at 25% below those of Qwen3.8-Max for both input and output tokens. The listing followed an earlier Qwen4-Max preview through Token Plan.
In September, Anthropic accused Alibaba of extracting Claude outputs. Anthropic later estimated that its newer defenses reduced exposed reasoning data by about 90% to 99%. That estimate does not establish an effect on Qwen4-Max's index score.
Qwen4-Max is the flagship of Alibaba's Qwen4 family. Alibaba describes it as designed for complex programming, front-end design and tasks that involve using tools across multiple steps. Its roadmap also includes Flash and Plus models and a 27B version for local deployment.